[TechSpot] Samsung's plans to kill the notch: Yes, please. - Page 4 - Overclock.net - An Overclocking Community

Forum Jump: 

[TechSpot] Samsung's plans to kill the notch: Yes, please.

Reply
 
Thread Tools
post #31 of 36 (permalink) Old 10-28-2018, 01:23 AM
New to Overclock.net
 
twitchyzero's Avatar
 
Join Date: Feb 2008
Location: Canada
Posts: 2,927
Rep: 82 (Unique: 62)
Quote: Originally Posted by Sir Beregond View Post

Really have never figured out how the curved edge screen was supposed to be of any value as an end user.
blame Apple and the social media generation

even electronics have to be sexy and aesthetically-pleasing

form over function

granted, I don't want my phone to look like it's from 2009, but brands like Samsung take it too far


no I don't care if it's 0.3mm thinner than last year, just give me something reliable, minimalistic and durable


isn't that one of the main causes that they found themselves with exploding Notes?

Nvidia Surround Gaming
My PC Hardware Upgrade CyclesWarning: Spoiler! (Click to show)
CPU: Athlon 1.3GHz | Core 2 Duo OC'd 2.0GHz | Core i5-3570K OC'd 4.8GHz
RAM: 2x256MB DDR-333MHz | 2x1GB DDR2-667MHz -> 2x2GB DDR2-800MHz | 4x4GB OC'd DDR3-2133MHz
GPU: GeForce2 MX 64MB -> Radeon 9600 Pro 256MB | Radeon X1950 Pro 256MB | GeForce GTX 670 FTW 2GB SLI OC'd

Last edited by twitchyzero; 10-28-2018 at 01:26 AM.
twitchyzero is offline  
Sponsored Links
Advertisement
 
post #32 of 36 (permalink) Old 10-29-2018, 09:02 AM
New to Overclock.net
 
Avonosac's Avatar
 
Join Date: Dec 2012
Location: PA
Posts: 2,957
Rep: 158 (Unique: 114)
Quote: Originally Posted by HybridCore View Post
Lens correction works differently because it's not guessing. It's remapping points to a new location to cancel out the effects of distortion in the lens. Neural networks make guesses and fill in information that didn't exist. They're very different beasts. One requires insane amounts of data to train the system and the other one is a much simpler set of transformations (albeit nonlinear).
Sigh, ok lets go.

It isn't remapping other points to new locations because those "points" have been condensed into other pixels. We aren't talking perfect lensing where pixel shape is a function of the lensing effect, we're talking dynamic occlusion. However, in either case you have a region of pixels which are inaccurate which need correction from (at least) the pixels around it. In this case the only difference between complete interpolation and 'adjustments' is the weight at which the pixels within the region are being trusted.

In the event of total occlusion, you would expect completely untrusted in the center and low trust on the border. While in a partial obstruction you're looking at a higher level of trust applied to the pixels within the corrected region.

The algorithm, even with a single source of input is exactly the same.

You'll note I also said they would use more than one sensor, which together would provide the similar "partially" trusted source data for the final composite image.


Avonosac is offline  
post #33 of 36 (permalink) Old 10-29-2018, 10:11 AM
New to Overclock.net
 
Join Date: Jan 2011
Posts: 10,002
Rep: 182 (Unique: 152)
Quote: Originally Posted by Avonosac View Post
Sigh, ok lets go.

It isn't remapping other points to new locations because those "points" have been condensed into other pixels. We aren't talking perfect lensing where pixel shape is a function of the lensing effect, we're talking dynamic occlusion. However, in either case you have a region of pixels which are inaccurate which need correction from (at least) the pixels around it. In this case the only difference between complete interpolation and 'adjustments' is the weight at which the pixels within the region are being trusted.

In the event of total occlusion, you would expect completely untrusted in the center and low trust on the border. While in a partial obstruction you're looking at a higher level of trust applied to the pixels within the corrected region.

The algorithm, even with a single source of input is exactly the same.

You'll note I also said they would use more than one sensor, which together would provide the similar "partially" trusted source data for the final composite image.
I'm not arguing that you're not losing resolution/information when doing software-based distortion correction compared to using actual corrective lenses. It's information that's permanently lost since it wasn't captured from the start. That still doesn't change the fact that software-based lens distortion correction is still just a remapping of information (basically stretching and squishing the input image) and AI a kind of trained guessing/filling algorithm in the simplest terms. They are still very different even if the end results are similar.

Last edited by HybridCore; 10-29-2018 at 10:30 AM.
HybridCore is offline  
Sponsored Links
Advertisement
 
post #34 of 36 (permalink) Old 10-29-2018, 12:13 PM
Web Developer full stack
 
charlesquik's Avatar
 
Join Date: Aug 2011
Location: quebec
Posts: 968
Rep: 23 (Unique: 21)
Take a look at Xiaomi Mi Mix 3.

There is a slider in the back to show the cameras.

Now this is a true no-notch phone.

charlesquik is online now  
post #35 of 36 (permalink) Old 10-29-2018, 01:46 PM
New to Overclock.net
 
Avonosac's Avatar
 
Join Date: Dec 2012
Location: PA
Posts: 2,957
Rep: 158 (Unique: 114)
Quote: Originally Posted by HybridCore View Post
Spoiler!
You have not understood my explanation of the algorithm at all, so you either aren't an engineer or you don't understand the fundamentals to both problems. If you don't understand software and _what_ AI is, I can't help you. But your summary of the "problem" very obviously demonstrates your lack of understanding.

Your assertion is false, these problems are the same thing.


Avonosac is offline  
post #36 of 36 (permalink) Old 10-29-2018, 11:01 PM
MegaTechPC
 
Majin SSJ Eric's Avatar
 
Join Date: Apr 2011
Location: Saint Simons Island, GA
Posts: 19,208
Rep: 1092 (Unique: 517)
I couldn't possibly care less about the supposed privacy implications of this, all I know is I WANT it!!!! Fully bezel-less screen up front with embedded cameras/sensors/etc is the holy grail of smart phone design and if Samsung pulls it off they will deservedly get the credit for having the proper vision of where these designs are going. Slow, cumbersome, and over-complicated motorized cameras are not the answer, embedded tech is where its at!


3DMark11 - P25138
3DMark Firestrike - P20998

Intel Core i7 4930K @ 4.7GHz | Asus Rampage IV Extreme | 2 x EVGA GTX Titan SC (1254MHz) | 16GB Patriot Viper Extreme DDR3 2133MHz (4 x 4GB) | Corsair AX1200 | Silverstone Temjin TJ11 | Corsair Force 3 240GB (System) | 2 x Intel 320 160GB SSD (Dedicated Gaming Drives) | Hitachi Deskstar 1TB (Data) | MS Windows 10 Pro | EK Supreme HF/FC-Titan/Rampage IV Extreme blocks | Hardware Labs GTX 560/240 rads | Alphacool VPP-655 D5 pump | Bitspower mod kit/pump top/fittings/120mm res


Majin SSJ Eric is offline  
Reply

Quick Reply
Message:
Options

Register Now

In order to be able to post messages on the Overclock.net - An Overclocking Community forums, you must first register.
Please enter your desired user name, your email address and other required details in the form below.
User Name:
If you do not want to register, fill this field only and the name will be used as user name for your post.
Password
Please enter a password for your user account. Note that passwords are case-sensitive.
Password:
Confirm Password:
Email Address
Please enter a valid email address for yourself.
Email Address:

Log-in



Currently Active Users Viewing This Thread: 1 (0 members and 1 guests)
 
Thread Tools
Show Printable Version Show Printable Version
Email this Page Email this Page


Forum Jump: 

Posting Rules  
You may post new threads
You may post replies
You may not post attachments
You may not edit your posts

BB code is On
Smilies are On
[IMG] code is On
HTML code is Off
Trackbacks are Off
Pingbacks are Off
Refbacks are Off